Every app on the App Store and Google Play carries a privacy summary on its store page: App Privacy on iPhone, Data safety on Android. For an AI app, look first at whether it collects your user content (your chats, photos and files), whether that data is linked to you, and whether anything is used for tracking or shared with other companies. An app whose AI runs on the phone should show little or no user content collected.
Both labels are filled in by the developer, so treat them as a declaration to check rather than a guarantee.
Where do you find the labels? #
- App Store: open the app’s page and scroll to App Privacy, then tap See Details for the full list.
- Google Play: open the app’s page and tap Data safety.
Check before installing. It takes a minute and usually tells you more than the app’s marketing copy.
How to read the App Store privacy label #
Apple groups data under three headings:
- Data Used to Track You. Linked with other companies’ data for targeted ads, or shared with data brokers. For an AI chat app, anything here deserves a close look.
- Data Linked to You. Collected and tied to your identity, such as an account, email or device ID.
- Data Not Linked to You. Collected but not tied to your identity.
An app can also declare Data Not Collected.
Within each heading Apple lists categories. The ones that matter for AI apps:
| Category | What it covers | Why it matters for AI |
|---|---|---|
| User Content | Emails or messages, photos or videos, audio, other user content | Your prompts, chats and attachments usually land here |
| Identifiers | User ID, device ID | Lets activity be tied together over time |
| Usage Data | Product interaction, advertising data | How you use the app; often from analytics or ad code |
| Diagnostics | Crash data, performance data | Usually harmless, but it’s still data leaving the phone |
| Contact Info | Name, email, phone number | Typically collected for accounts |
| Sensitive Info | Racial or ethnic data, sexual orientation, religious beliefs and similar | Should almost never appear for an AI chat app |
How to read Google Play’s Data safety section #
Google’s layout is different. It shows:
- Data shared: sent to other companies or organizations.
- Data collected: gathered by the app itself.
- Security practices: whether data is encrypted in transit, and whether you can request deletion.
The types to watch for an AI app are Messages, Files and docs, Photos and videos, Audio files, App activity and Device or other IDs. If an AI app lists Messages or Files and docs as collected and shared, your prompts and attachments are leaving the phone.
What does a good label look like for an AI chat app? #
- Cloud AI app: User Content will be collected, because the server has to receive your messages in order to answer. The real questions are whether it’s linked to you, used for tracking, or shared. Then check the training and retention settings inside the app.
- Local AI app: User Content shouldn’t be collected at all, since the model runs on your phone. Expect an ad-supported one to list what its ad network collects, such as identifiers or usage data, and possibly diagnostics.
What labels don’t tell you #
- Whether your chats train future models. Training use isn’t a label category anywhere. You need the privacy policy and the in-app settings. Do humans read your AI chats covers what each big provider does.
- How long data is kept. Retention periods live in the privacy policy.
- Human review. Whether staff can read samples of your conversations isn’t shown.
- Accuracy. Developers complete the labels themselves. Apple and Google have policies and can remove apps that misreport, but they don’t audit every label.
Other platform rules that now apply to AI apps #
Store rules have tightened around AI specifically. Apple’s App Review Guidelines require apps to “clearly disclose where personal data will be shared with third parties, including with third-party AI, and obtain explicit permission before doing so” (guideline 5.1.2(i)). So an iPhone app that pipes your text to an outside AI provider should tell you and ask first.
Apple also asks developers to declare what common third-party SDKs collect through privacy manifests, which feeds into more accurate labels. And apps must ask through App Tracking Transparency before tracking you across other companies’ apps and websites. Separately from store rules, disclosure laws are arriving for chatbots themselves; see do AI chatbots have to tell you they’re AI.
Red flags on an AI app’s label #
- User Content under “Data Used to Track You.” Your chats or photos are feeding advertising.
- User Content shared with third parties on Google Play, with no explanation anywhere in the app.
- Sensitive Info collected by a general-purpose chatbot.
- A label that contradicts the pitch, such as “100% private, on-device” alongside User Content linked to you.
- No deletion option under Google Play’s security practices.
How to double-check a label #
Read the privacy policy’s AI section, searching for “train,” “retain,” “third party” and “AI provider.” Try the app in airplane mode, since a local one keeps working. On iPhone, turn on App Privacy Report under Settings, Privacy & Security to see which domains the app actually contacts. And watch for the third-party AI permission prompt: if an app sends your data to another AI company, it should ask. The full set of checks is in how to tell if an AI app is really private.
Applying it to Personal LLM #
Personal LLM runs its models on your phone, so your chats, photos and prompts never leave it. There’s no account, no cloud sync and no analytics on conversations. The free version shows Google AdMob ads, and AdMob may collect device information to serve them, as the privacy policy explains. That’s the kind of entry you’d expect to see declared for an ad-supported app, and it’s why an ad-supported free app shouldn’t be described as 100% private while you’re online. On iOS, declining tracking keeps the ads non-personalized, and a one-time purchase removes them. Compare its current store label against that description before you install. For how ads and AI chats intersect across the industry, see do AI chatbots use your chats for ads?.
Frequently asked questions #
What are App Store privacy labels? #
Summaries on each App Store page, under App Privacy, of the data an app collects, whether it’s linked to you, and whether it’s used to track you. Developers provide the information themselves.
What is the Google Play Data safety section? #
Google Play’s equivalent of a privacy label. It lists the data an app collects and shares, whether data is encrypted in transit, and whether you can ask for it to be deleted.
Are app privacy labels accurate? #
Usually roughly right, but not always. They’re self-reported and the stores don’t verify each one, so use them alongside the privacy policy and simple tests such as trying the app in airplane mode.
What should an AI app’s privacy label show? #
A local AI app should show no user content collected. A cloud AI app will collect your messages, so check that they aren’t used for tracking or shared, then review its training and retention settings.